The clustering of high schools based on national and school examinations: A case study at Daerah Istimewa Yogyakarta Province

The purpose of Indonesian National Examination for High School Students is to measure and assess students' knowledge and competence in particular subjects. The result is also going to be used as one of consideration for mapping Indonesia's national education quality. Aside from National Examination (NE), each school also conducts School Examination (SE). Both examinations are supposed to represent quality of education since the examinations measure the competence of the same students. However, the results of both examinations are not always linear [1]. In fact, the need of NE in Indonesian education is still being pro and cons among society. In order to identify whether NE and SE could be used to represent the quality of educations in Daerah Istimewa Yogyakarta Province, this paper describes the analysis of NE and SE score by performing data mining technique using Fuzzy C-Means clustering algorithm towards NE score and SE score independently. Furthermore, the clusters were analyzed using univariate Anova, Spearman correlation, and crosstabulation. The data used in this research are NE and SE scores of Natural Science Department and Social Science Department of all high schools in Daerah Istimewa Yogyakarta Province from academic year 2011/2012 to 2014/2015. The results of cluster analysis are three different clusters of NE in Natural Science Department, three different clusters of NE in Social Science Department, three different clusters of SE in Natural Science Department, and three different clusters of SE in Social Science Department for each year. The clusters are significantly separated. There is an opposite direction relationship between clusters of NE and SE. The relationship is weak which means there is no guarantee that a school which belongs to cluster-i of NE will be in the cluster-i of SE. Both for NE and SE memberships, only few members migrated from one cluster to another across years. The number of schools having the same cluster of NE and SE in each department varies from year to year, but generally less than 22%. The migrations of NE and SE cluster members from higher cluster to lower one and vice versa also vary. In addition, there is inconsistency clustering based on NE and SE. Since SE is not standardized and indeed is a formative test, there might be subjective aspects involved in grading the students. Therefore, if the government intends to map Indonesia's national education quality, national examination is more suitable than school examination for this purpose.